How To Implement The Health Insurance And Reinsurance Programs In New York

Transcription

1 Risk Adjustment and Reinsurance under the ACA New York State Recommendations Prepared by Wakely Consulting Group Ross Winkelman, FSA, MAAA and Syed Mehmud, ASA, MAAA With contributions from: Mary Hegemann, FSA, MAAA; James Woolman; Julie Peper, FSA, MAAA; Patrick Holland For more information on any of the information contained in this paper, please contact Ross Winkelman, FSA, MAAA at or (720) June 2012 Support for this resource was provided by a grant from the Robert Wood Johnson Foundation s and Funding from the New York State Health Foundation

2 Contents 1 Executive Overview Introduction to Risk Adjustment, Reinsurance and Risk Corridors Key Decisions Program Responsibility Federal or State Risk Adjustment Model and Key Technical Issues Federal or State Reinsurance Parameters Level of Stakeholder Engagement Other Structural Exchange Decisions Simulations and Other Preparation Different Approaches for Simulations Key Simulation Deliverables Filing with HHS Administration and Governance Determine Program Governance and Oversight Program Financing Establish Administrative Infrastructure Risk Adjustment Develop Administrative Infrastructure Reinsurance Establish Funds Flow Mechanisms and Cash Management Plan Develop Reporting and Transparency Plan Establish Data Review and Audit Program Coordination with MLR, Risk Corridor, and Other ACA Provisions Support for this resource was provided by a grant from the Robert Wood Johnson Foundation s State Health Reform Assistance Network Program and Funding from the New York State Health Foundation

3 1 Executive Overview On July 11, 2011, the U.S. Department of Health and Human Services (HHS) issued proposed rules, titled Patient Protection and Affordable Care Act; Standards Related to Reinsurance, Risk Corridors and Risk Adjustment. On March 16, 2012, HHS issued final rules. The rules implement standards for these programs for states and health insurance issuers (e.g. health insurance companies and HMOs). These programs are intended to mitigate the impact of adverse selection and lessen the financial risk health insurance issuers ( issuers ) will face under the Affordable Care Act (ACA). Under separate cover titled, Analysis of HHS Proposed Rules on Reinsurance, Risk Corridors, and Risk Adjustment, 1 Wakely has already provided a summary of the proposed rules and resulting implications, and will soon be releasing a revised version on the final rules. Wakely has provided support to the state of New York through a series of meetings and stakeholder outreach efforts. These efforts were supported through grants from the New York State Health Foundation (NYSHealth) and the Robert Wood Johnson Foundation (RWJ). This report is the culmination of the efforts to date by Wakely, NYSHealth, RWJ, the NY Department of Health (DOH), the NY Department of Financial Services (DFS), individuals at CCIIO, and representatives from the health plans. This paper provides recommendations for New York s approach to addressing these requirements and implementing these programs, along with the rationale for the recommendations. Pending federal and state regulation and guidance, and insurance market structural decisions made by the state of New York may materially affect the recommendations or timelines in this paper. Our recommendations focus on the areas where we have the most informed perspective and the ones that are most urgent. The final design and implementation of these programs will require decisions in other areas, and more detailed planning than is covered by this report, especially related to operational and governance issues. The opinions presented in this report are those of the authors, not others at Wakely or within the State of New York or a single health insurance issuer. This report is for purposes of informing the direction of these programs in the State of New York and may not be appropriate for other purposes. Additional Federal and State rules may materially affect the recommendations contained in this report. Wakely does not intend to create a duty or liability to any outside party. 1 Wakely Consulting Group, Analysis of HHS Proposed Rules On Reinsurance, Risk Corridors And Risk Adjustment, September 2011, Page 1

4 New York currently has several risk adjustment and reinsurance programs in place now or which have been suspended but are still relevant, summarized in the table below: Current Commercial NY Risk Adjustment and Reinsurance Programs Compared to Post Reform Reg 146-4th Amendment (Risk Reg 146-5th Amendment Reg 171 (Reinsurance ) Reg 171 (HealthyNY) Post 2014 (ACA) - Risk Adjustment Post 2014 (ACA) Reinsurance Post 2014 (ACA) - Risk Corridor Current Market Adjustment) 1 (High Cost HIX Only) Definition Claims %) Direct Pay HMO X X X X X X Direct Pay POS X X X X X X Direct Pay Other X X X X X Healthy NY Individual X X X X Healthy NY Small Group X X X Other Small Group X X X X 1 The Reg 146, 4 th amendment risk adjustment program is no longer in place, having been replaced by the high cost claims % program. It is included because it more closely resembles the Post Reform risk adjustment program than the high cost claims % program. The New York Medicaid managed care program also has used risk adjustment for a number of years. The Medicaid risk adjustment program, which is discussed in more detail later in this paper, uses CRGs. The model currently used for the Medicaid program is closest to the Reg th Amendment program because diagnosis based condition categories are used, rather than actual expenditures like the 5 th Amendment and Reg 171 programs. As shown above, these programs apply to markets differently than the programs required under reform. Also, the current programs do not comply with other requirements of the reform programs 2. In addition to the programs above, New York also has a risk adjustment program for Medicare Supplement policies. Throughout this paper, when the phrase current reinsurance programs is used, it refers to the Reg 146 5th amendment and Reg 171 programs. The Reg 146, 4 th amendment program is no longer in place but was included in the table above because it was a claims based risk adjustment approach which applied to the commercial market (and therefore is most similar to the ACA risk adjustment program). In 2010, Reg 146 payments were about $146 Million and Reg 171 payments were about $63 Million. Reinsurance payments are expected to be approximately $600 Million in 2014 under the ACA reinsurance program, although this amount is subject to significant uncertainty. 2 The ACA includes very specific requirements for these programs including the structure and performance of the risk adjustment model, the structure of the reinsurance parameters, the markets the provisions apply to and in the case of reinsurance, the minimum amount of coverage that needs to be offered. Page 2

5 We worked with the state of New York and the New York State Health Foundation (NYSHealth) to engage CCIIO, issuers and state health program policymakers and administrators in discussions related to the Risk Adjustment and Reinsurance programs under the ACA, and specifically the methodology the State of New York should use. Those discussions included a large group policy meeting with CCIIO and state personnel, a large group technical meeting with state personnel and CFOs and actuaries from issuers, and one on one, confidential phone calls between Wakely consultants and issuers. Based on these stakeholder outreach efforts, phone calls and other meetings with state personnel, and professional judgment, we have the following recommendations for New York implementation of the ACA s risk adjustment and reinsurance provisions: 1. Because of the state of New York s market size, rating rules, geographic and issuer variation, its experience working on risk mitigation programs in the commercial and Medicaid markets, available grant funding to create an all payer claims database (APCD) and other issues to create a state specific solution, we recommend that the state administer the Reinsurance and Risk Adjustment programs. The health plans generally supported a state based approach in the stakeholder engagement meetings held to date. We would expect the New York Department of Health (DOH) and the New York Department of Financial Services (DFS) to have roles in the programs. DOH is adept at handling detailed encounter data and running a commercially available risk adjustment model (CRGs from 3M). DFS has substantial experience administering risk mitigation programs, collecting assessments, reviewing rate filings in the commercial market, and auditing health plan data submissions. Under the federal requirements, administration of the programs could reside at DOH, DFS, or a new agency, sub-agency or non-profit, non-governmental organization. 2. Ultimately, we recommend collection of detailed data by the State rather than the distributed approach where health plans run the risk adjustment model and provide summary level information back to the State (HHS will be using a distributed approach where they administer risk adjustment on behalf of states according to the final rules which were just recently released). However, New York s APCD will likely not be ready for use in Therefore, New York risk adjustment in 2014 should follow a distributed approach. 3. We recommend use of CRGs. CRGs are familiar in New York and the DOH has had worked very successfully with 3M (the developers of CRGs) in the past. If contracting or other issues do not allow use of CRGs, the federal model will likely be the best alternative option. 4. The Reg th Amendment and Reg 171 risk adjustment and reinsurance programs should be eliminated as of 1/1/2014 since they do not comply with HHS final rules and would have to be fundamentally altered to do so. In addition, they would not be complementary to the ACA required programs. Finally, the ACA programs will provide Page 3

6 more objective and comprehensive protections than these programs. In 2010, Reg 146 payments were about $146 Million and Reg 171 payments were about $63 Million. Reinsurance payments are expected to be approximately $600 Million in 2014 under the ACA reinsurance program, although this amount is subject to significant uncertainty. 5. The State needs to start the risk adjustment and reinsurance simulation and analysis process soon, and no later than early July 2012, with significant engagement by issuers. Data quality will be the key driver for the success or failure of New York s (and any state s) risk adjustment program. Analyzing reinsurance parameters will also be crucial in this exercise in order to be ready to publish modifications in the state notice of benefit and payment parameters by March 1, Timing considerations cannot be overemphasized in the design of these programs. New York should carefully consider the required timing for health plan pricing and valuation, and the interplay between risk adjustment, reinsurance, risk corridors and MLR requirements when designing the risk adjustment methodology and the administrative structure for both risk adjustment and reinsurance. More detail on these recommendations and the rationale behind them are provided in the remainder of this paper. Our recommendations represent out best judgment with respect to these design decisions. However, in our opinion, other options represent reasonable approaches in many cases. 2 Introduction to Risk Adjustment, Reinsurance and Risk Corridors Risk Adjustment in the context of this paper is a term used to describe methods of adjusting premium rates for differences in the underlying morbidity of a health plan s membership. Morbidity is measured using sophisticated tools, called risk adjustment software. These tools use some combination of information including diagnosis data, demographic (age and gender) information, and the types of prescription drugs that someone may be taking to estimate the morbidity of that person. Actual costs are not used to measure or predict morbidity although they are used to develop relative payment weights for severity of illness ( case mix ) calculations. Page 4

7 The following example illustrates a regression model approach 3 to developing a risk score for a single member: Risk Marker Risk Weight Male, Age Diabetes with significant co-morbidities 1.32 Asthma/COPD 0.96 Low cost dermatology 0.30 Total Risk Score 2.80 The risk score for a particular individual is the sum of the weights associated with all conditions that the data indicate the person has (2.80 in the above example). Diagnosis codes and/or pharmacy national drug codes (NDCs) map to the condition categories. The ACA Risk adjustment is a permanent program (i.e. will continue indefinitely as opposed to reinsurance and risk corridors which are only in effect from 2014 through 2016), applying to the individual and small group markets. Reinsurance in the context of this paper is a term used to describe methods for protecting health plans against high cost individuals. Reinsurance methods usually pay back health insurance companies for some portion (called coinsurance) of costs above some level (called the attachment point). Sometimes, there is a maximum reimbursement amount or maximum amount to which the reinsurance will apply, as is the case in the ACA s reinsurance program. The following example highlights the structure of the ACA s reinsurance program and how it differs from typical commercial reinsurance. 3 The Federal model will be a regression based approach, while CRGs are categorical. Categorical models assign members into a discrete number of descriptive categories that are not specific to a single condition. Regression models flag each and every condition a member has (although they are usually hierarchal, meaning only the most severe condition within a category of conditions is flagged). Page 5

8 Sample Reinsurance Calculation Reinsurance Parameters State or Federal Reinsurance Traditional Reinsurance Attachment Point (paid claims threshold where reinsurance begins) $50,000 $200,000 Coinsurance Rate (percent between attachment point and cap for which reinsurer is liable) 80% 85% Reinsurance Cap (claims in excess of the cap are not eligible for reinsurance) $150,000 $2,000,000 Example Insurer Initial Paid Claim Amount = $500,000 Net Insurer Liability* = $50, % x (150,000-50,000) + (200, ,000) + 15% x (500, ,000) = $165,000 State or Federal Reinsurance Payment* = 80% x (150,000-50,000) = $80,000 Traditional Reinsurance Payment = 85% x (500, ,000) = $255,000 * Note that the State/Federal Payments may be prorated down for all insurers if the total payments exceed the available funds The ACA reinsurance provisions are a little different than typical commercial reinsurance in that the attachment point and cap are relatively low. This is because the temporary ACA program was designed to complement, not replace commercial reinsurance. The ACA Reinsurance program begins in 2014 and is scheduled to end in It applies to the individual market. Risk corridors in the context of this paper protect health plans against unfavorable financial results, and also prevent health plans from experiencing unreasonably favorable results. There is a corridor around the target loss ratio within which the health plan is at risk for favorable or unfavorable results (3% under the ACA). In excess of this 3% corridor, the health plan and the federal government share in the results. The ACA Risk Corridor program begins in 2014 and is scheduled to end in It applies to Qualified Health Plans (QHPs) offered in the individual and small group markets (primarily those plans offered through the Health Insurance Exchange). This paper is focused on the risk adjustment and reinsurance provisions of the ACA because these programs have state specific elements and state options for administering the programs, while the risk corridor program is federally administered with no state specific options. However, it is important for the state to understand the risk corridor program and integrate other programs, legislation and structural decisions, including the risk adjustment and reinsurance programs with its provisions. Page 6

9 3 Key Decisions States need to make a number of key decisions with respect to the risk adjustment and reinsurance provisions of the ACA. Decision areas that fundamentally affect the level of effort and timing of these programs are included below, listed roughly in order of importance. A full discussion of each decision, including the detailed technical issues is outside the scope of this work plan. Our input focuses on major design decisions and the most urgent issues. 3.1 Program Responsibility The final rules allow states to manage their reinsurance program if the state is also operating an exchange or to defer operation to HHS (the option to defer to HHS is a change from the proposed to the final rules). If a state is not operating an exchange, they can still manage the reinsurance program or let HHS administer it. Risk adjustment is similar in that local operation of an exchange allows that state to also utilize a methodology different than the federally prescribed one. States may also allow HHS to administer the risk adjustment program even if they have a state-based exchange. The table below reflects the final rules for purposes of clarifying the various options available and to reinforce the markets to which each program applies. Sold within Exchange Sold Outside Exchange Who Administers ACA Provision Individual Small Group Individual Small Group Grand- Fathered State Run Exchange Federal Run Exchange Risk Adjustment Yes Yes Yes Yes No State or HHS * HHS Reinsurance Yes No Yes No No State or HHS * State or HHS * Risk Corridor Yes Yes Some ** Some ** No HHS HHS * State can decide to administer or allow HHS to administer. If HHS administers, all parameters will be federal **Risk Corridor will apply to QHPs sold outside the HIX that are substantially the same as those sold within the HIX There are a number of important issues that go into the decision to manage one or both of these programs. The level of resources required to administer these complex programs and the availability of data (i.e. through an All Payer Claims Database [APCD]) are very significant issues, especially for risk adjustment. In addition, the state s (and other stakeholders ) desire to control these programs, particularly risk adjustment, may drive the state toward taking on this responsibility. If New York decides to have HHS administer one or both of the reinsurance and risk adjustment programs, work required in 2014 and beyond decreases considerably. However, the work in 2012, and 2013 does not change significantly since issuers need information on the impact of risk adjustment and reinsurance to develop pricing for 2014, and HHS will not be able to provide significantly detailed information prior to The resources necessary to manage reinsurance are lower than those required to manage risk adjustment, but are still significant. Page 7

10 Recommendation: Because of New York s market size, rating rules, geographic and issuer variation, its experience working on risk mitigation programs in the commercial and Medicaid markets, available grant funding to create an all payer claims database (APCD) and other issues, we recommend that the state administer the Reinsurance and Risk Adjustment programs. The health plans generally supported a state based approach in the stakeholder engagement meetings held to date. 3.2 Federal or State Risk Adjustment Model and Key Technical Issues Risk adjustment programs require a risk adjustment model. The proposed rules indicate that HHS will release a federal model in October States that want to use an alternative model 4 need to submit it to HHS for review in November 2012, with HHS proposing a maximum two-month turnaround for their review (by January 2013). The short time period between the release of the federal model and required submission of state models means that states need to start analyzing alternative models early in 2012 if they are at all considering an alternative model. Other hybrid options include using the federal model, but making fundamental changes to it like recalibrating the risk weights. Per the final rules, HHS will require states to use the national approach to calculating payments and charges (released in October 2012), at least in the initial years of the ACA. They may allow states operating the risk adjustment program to use a state alternative approach to the calculation of payments and charges in the future to foster innovation. Therefore, if New York decides to pursue an alternative risk adjustment program, they will only be able to modify data collection, the choice of the risk adjustment model, and the calculation of plan average actuarial risk. Besides the basic choices of model and data collection approach, the following key technical decisions will need to be made if New York administers risk adjustment: a) Prospective vs. Concurrent/Retrospective model b) Consideration of metallic tiers c) Include pharmacy categories or not d) Data fields to be used (e.g. first five diagnosis fields versus all available) e) Rating variables and rating variable integration f) Area calculations and adjustments g) Scoring for members with limited experience h) Whether to phase-in the application of risk adjustment i) Whether to adjust the model parameters for reinsurance payments 4 A number of risk adjustment models are currently being used for risk-adjusted payment in Medicare, Medicaid and other public programs including Medicare s HCC, CDPS, MedicaidRx, ACGs, ERGs, and DxCG. Others have been developed specifically for reform programs including Milliman s MARA, Johns Hopkins ACG reform model, and Wakely s WRA model. Page 8

11 The types of data used is likely a more important decision than which software model to use. All models use demographic data and with that diagnosis data, pharmacy data or both pharmacy and diagnosis data. There are a number of important data characteristics to consider in selecting a model that uses diagnosis data, pharmacy data or both: Diagnosis Data 1. More difficult to collect 2. Data quality can be a problem for some organizations and differences in data quality can drive some of the differences in risk scores 3. May take several months to be reported to health plan 4. Diagnosis codes are relatively stable (major updates don t frequently happen. The ICD- 10 conversion may happen in , but this should not create any major challenges because the ICD-10 s map fairly easily to ICD-9 s) 5. More differentiation within condition categories, especially for a concurrent model which is recommended for 2014 and May incentivize providers to record diagnoses that are more severe than the actual condition the patient has 7. Diagnosis only risk adjustment models have shown good statistical performance, although real world conditions hamper their performance more than pharmacy models. Their statistical performance increases significantly compared to pharmacy only models under concurrent application. 8. Use of diagnoses to develop risk adjustment categories allows model builders more discretion in defining condition categories for payment than pharmacy only models allow. This makes it easier to develop a risk adjustment system that incentivizes efficient care delivery. Pharmacy Data 1. Easy to collect 2. Relatively uniform quality across organizations as long as pharmacy benefit is provided uniformly 3. Reported quickly to health plan after the prescription is filled 4. Pharmaceuticals change relatively quickly (new drugs or formularies) 5 Was originally scheduled to occur in late 2013, but has been delayed. Page 9

12 5. May not differentiate well within disease categories 6. May incentivize health plans or providers to prescribe drugs that will not benefit and may harm patients in order to maximize revenue. However, given that the pharmacy only approach would only be used in transition, it would be unlikely to create meaningful incentives for health plans or providers to behave in detrimental ways. 7. Off label uses may create false indications for some conditions 8. People tend to use prescription drugs on a regular basis, so the underlying condition shows up relatively consistently in the data 9. Pharmacy only risk adjustment models have shown good statistical performance under real world conditions (similar or better for prospective applications, but worse for concurrent applications) 10. Use of pharmacy only model does not allow model builders much discretion in defining condition categories for payment. Therefore, the use of a pharmacy only model makes it more difficult (or impossible) to develop a risk adjustment system that incentivizes efficient care delivery. Of all of the issues above, one of the primary concerns is the quality of data across all organizations. If New York were going to collect detailed data from health plans and use that data to develop plan average actuarial risk factors, we would recommend a pharmacy data only approach in 2014 and possibly 2015 given the current state of the APCD. However, if New York uses a distributed approach initially, we would recommend use of full data (diagnosis and pharmacy) given the advantages of diagnosis data listed above. Given that HHS will be using a distributed approach to risk adjustment and the status of the New York APCD, we recommend the distributed model approach is used in 2014 and likely New York currently uses the Clinical Risk Groups risk adjustment model, developed by 3M, for Medicaid managed care risk adjustment. We are not aware of other vendor models that are currently used within New York government programs. It is likely that some health plans license other risk adjustment models for internal disease management, budgeting or other purposes. If New York decides not to use a pharmacy only model initially and CRGs is not a viable option because of contracting or other issues, use of the Federal Risk Adjustment model is a potential option. The most important characteristics of model choice are transparency, familiarity, data used, and cost. Technical performance (measured by Individual level R-squared or other statistical measurements) is relevant, but lower on the list of criteria and most models perform similarly. Page 10

13 Recommendation: Ultimately, we recommend collection of detailed data by the State rather than the distributed approach where health plans run the risk adjustment model and provide summary level information back to the State (HHS will be using a distributed approach where they administer risk adjustment on behalf of states according to the final rules which were just recently released). However, New York s APCD will likely not be ready for use in Therefore, New York risk adjustment in 2014 and possibly 2015 should follow a distributed approach. We recommend use of either the CRG model or the federal risk adjustment model. Risk adjustment simulations should focus on data quality to identify and correct any major problems across issuers. Issuers should be allowed and encouraged to participate in the simulation process. As methodology options are considered and analyzed, only one set of results should be shared with health plans to ensure stakeholder input is constructive rather than focused on optimizing financial results. 3.3 Federal or State Reinsurance Parameters States can use the federal reinsurance parameters or develop state-based parameters. The minimum contribution rate (what all issuers and TPAs will contribute to fund reinsurance) will be set uniformly on a national basis. HHS is expected to publish federal reinsurance parameters based on the market characteristics of each state rather than publishing only one set of federal parameters to be applied to all states. The primary issue that will drive each state s reinsurance parameters is the size of the projected non-group market compared to the overall insurance market to which assessments apply in 2014, 2015 and However, some states may have an expected mix of healthy and sick enrollees in the individual market that is different than the national average or may be assumed by HHS. In addition to the federal contribution rate, key reinsurance parameters that will be defined by HHS around mid-october of 2012 are as follows: a) Attachment point b) Reinsurance cap c) Coinsurance rate The following example, repeated from the Introduction section, shows an illustrative attachment point, reinsurance cap and coinsurance rate and how these values affect he net insurer liability: 6 The assessments will apply to fully insured commercial business and self-funded employer groups. Page 11

14 Sample Reinsurance Calculation Reinsurance Parameters State or Federal Reinsurance Traditional Reinsurance Attachment Point (paid claims threshold where reinsurance begins) $50,000 $200,000 Coinsurance Rate (percent between attachment point and cap for which reinsurer is liable) 80% 85% Reinsurance Cap (claims in excess of the cap are not eligible for reinsurance) $150,000 $2,000,000 Example Insurer Initial Paid Claim Amount = $500,000 Net Insurer Liability* = $50, % x (150,000-50,000) + (200, ,000) + 15% x (500, ,000) = $165,000 State or Federal Reinsurance Payment* = 80% x (150,000-50,000) = $80,000 Traditional Reinsurance Payment = 85% x (500, ,000) = $255,000 * Note that the State/Federal Payments may be prorated down for all insurers if the total payments exceed the available funds States wanting to file either contribution rates or parameters different than the federal ones must respond to HHS by November 2012 for the 2014 benefit year. Final notice of federal factors will be in January If New York alters any parameters, public notice needs to be provided no later than March 2013 for use in While not a reinsurance parameter per se, however as part of analyzing the parameters the State should consider whether it wants to collect additional amounts than those that would be collected based on the national contribution rate set by HHS (and published in Oct 2012). The additional amount may be used towards (a) funding for administrative expenses of the applicable reinsurance entity; or (b) additional funding for reinsurance payments. Recommendation: New York should develop state specific reinsurance parameters and then compare these to the federal parameters for New York once the federal parameters are released. We would expect New York to file state specific values because New York will be able to do more detail and careful analysis specific to the state of New York than HHS will be able to perform. The process of comparing New York parameters developed by the state to the New York parameters developed by HHS may help inform the state specific values, suggest additional analysis is necessary, or support the developed values. Additionally, if New York decides to administer the program, it should test whether additional amounts need to be collected in order to further decrease premiums for the non-group market post reform or for the administration of the program. Page 12

15 3.4 Level of Stakeholder Engagement The stakeholder engagement process in New York has included three important outreach efforts to date: 1. Policy Meeting On December 7, 2011, the NYS Health Foundation in conjunction with the state, 2. Technical Meeting On December 8, 2011, the NYS Health Foundation in conjunction with the state 3. Individual Meetings Between November 2011 and January 2012, Wakely met via phone individually with nine health insurance companies to discuss important issues related to these programs and specifically the recommendations The goals of the discussions were presented as informing Risk Adjustment and Reinsurance program development in New York under ACA. The following issues were discussed (in bold) with summary of the input provided by health plans collectively immediately following each item: 1. What important issues / concerns do you think need to be addressed specific to their plan or the NY market The main concerns that were expressed were the uncertainty associated with the impact of the programs and what happens to the existing programs. There was widespread support for an accelerated process to begin simulations and technical analysis, decision making, and further stakeholder engagement. 2. Federal or state model (software) Most health plans assumed that the state would move towards a state based approach. This included both the model and methodology. There was some support for a Federal model and methodology, but more support for a state approach. During follow up discussions more recently, some preference was stated for a Federal approach. 3. Federal or state methodology Plans had the same reaction as Federal or state model decision health plans generally assumed that New York would choose to administer these programs. Factors discussed included the timing resulting from either a Federal or State approach, and whether the state had the resources necessary to administer the program. The final rules do not allow states the option of using a state specific approach to calculating payments and charges within the risk adjustment methodology, only in calculating average actuarial risk by health plan and integrating rating variables. Page 13

16 4. Any perspective on administration of the programs DFS, DOH, other Opinions on this issue varied. Many health plans assumed DFS would administer programs like it currently does under Reg146 and did not express any concerns. The only concern expressed related to the timing of payouts for the new programs because of timing issues under the current programs. Federal timing requirements would seem to address concerns here and the State will need to ensure compliance. 5. How important is integration with Medicaid There was general acknowledgement that the approaches would necessarily be different. However, also general support for as much consistency as possible between the ACA and Medicaid approaches to risk adjustment. Some of the health plans did not participate in Medicaid and, therefore, did not have an opinion. 6. If state, what model / methodology The two choices preferred by health plans were CRG and the federal model. The general view was that understanding the details of a given methodology and how it would be applied was more important than the choice of a model. Most plans expressed that transparency of the selected model and methodology was important to them. 7. How important are timing considerations for both pricing and valuation More than any other topic, this was a concern for the health plans. There was widespread recognition of the challenges they would be facing to price their products in 2013 for There was not a strong sentiment relating to cash-flow issues with settling risk adjustment payments in Most plans however expressed significant concerns with not knowing ultimate earnings until Some plans supported more frequent and interim risk score calculations to inform estimates of ultimate revenue. Plans also expressed a desire to know the risk adjustment and reinsurance rules and methodology by December 2012 so that they can begin to incorporate the impact into their pricing. 8. Trade-off between robust program and assessments to pay for administration This was not the focus of much of the discussion. The health plans generally preferred a less cumbersome process wherever possible and seemed to prefer less overhead to a program that was more precise and may protect some plans better. 9. What about audit program robust or low cost / non-invasive There was widespread agreement that the audit program should satisfy federal requirements, and try to identify true outliers, but should be as limited as possible. Page 14

17 10. Reinsurance do they have outside reinsurance for their individual business? If so, what type and what are attachment points? Most health plans had commercial reinsurance. Attachment points varied from $100,000 to $1,000,000. However, most had attachment points between $200,000 to $300,000 which is consistent with work we have completed in other markets. 11. Would they prefer state is aggressive in setting reinsurance parameters (aggressive means chance assessments wouldn t be enough to cover payouts)? There were not uniform answers to this question other than preferring that the estimates were consistent with actual results and that significant shortfalls did not occur. Some of this perspective related to current New York risk programs. 12. Is it preferable to extend reinsurance beyond 2016 or cover as much as possible from 2014 to 2016? This was not a major area of concern for health plans. 13. What information do you want to have in 2013 to price in 2014? The typical information was requested risk scores, impact of reinsurance, and some estimate / information on the impact of the previously uninsured. 14. Level of interest in participating in a risk adjustment simulation in advance of 2014 to inform 2014 pricing? There was widespread support for simulations and no important concerns were expressed with participating in such a simulation. Recommendation: New York should continue the stakeholder engagement process given the strong interest and constructive approach from the health plans. New York should continue to solicit feedback on methodological issues before providing health plan specific results to the health plans. Methodological decisions should be made prior to providing results to health plans to ensure a continued constructive input process. 3.5 Other Structural Exchange Decisions A number of other state-delegated decisions may affect the reinsurance and risk adjustment programs including the following: Will the state operate a Basic Health Program? Will the non-group and small group markets be merged? Will employer groups of be included in small group prior to 2016? What employer options will be allowed within the SHOP Exchange? Page 15

18 A full discussion of the technical impacts of the above decisions to the risk adjustment and reinsurance programs is outside the scope of this paper. However, it is important that the risk adjustment and reinsurance approaches ultimately implemented consider the implications of these structural decisions. Our recommendations may change depending on the decisions made surrounding these issues. Recommendation: As key structural decisions are made, the State should revisit the recommended approach for risk adjustment and reinsurance. 4 Simulations and Other Preparation New York issuers will be faced with significant uncertainty with respect to pricing their products in 2014 because of the significant changes under the ACA, including the impact of the risk adjustment and reinsurance programs. Where issuers experience uncertainty, they often increase premium rates or simply choose not to offer products in that market, either of which will protect their organizations from exposure to excessive financial risk. Both reactions are undesirable from a state perspective. No matter how much analysis is completed prior to 2014, significant uncertainty will still exist. However, analysis can be performed which will lessen the uncertainty associated with the risk adjustment and reinsurance programs. From an actuarial perspective, this analysis may be necessary for actuaries to issue unqualified rate certifications that will comply with actuarial standards of practice. States and issuers will need to work together to effectively analyze options, make decisions and simulate the impact of various methodologies. New York pricing actuaries will need three key pieces of information with respect to the risk adjustment program in order to price their products for 2014: 1. What is the risk score of their current enrolled population with respect to the market average? 2. What is the average cost of a currently uninsured individual who is expected to join the insurance market relative to individuals currently insured, and what proportion of the total market in 2014 will these members represent? 3. What is the risk score of the currently uninsured population that will become insured in 2014 (most relevant for the non-group market)? In the above questions, risk score and average cost are defined as relative to average members in their rating category (i.e. after accounting for allowable rating variables such as age, smoking status, and geographic area). However, it is expected that New York will retain pure community rating so these rating variables are not allowed and the calculations are therefore simplified. Page 16

19 Question 2 above is typically included in Level One Establishment Grant proposed activities and is addressed by economists and/or actuarial consultants. The Urban Institute has completed work related to question 3 above 7 for the state of New York. However, their report does not address Question 3. In addition to the questions related to the risk adjustment program above, health insurance issuer pricing actuaries will need to develop an estimate of the impact of the reinsurance program on non-group product pricing. The state will need to perform careful modeling to determine appropriate reinsurance parameters. If the parameters are set too conservatively or aggressively, the reinsurance program may end up with excessive reserves or shortfalls, either of which could be detrimental to an efficient market. Other market information will be relevant to simulations and program development, including the following: Overall market share for each market (non-group, small group, fully insured group, ASO/TPA, Association, and Medicaid), and by health insurance issuer. Currently uninsured and changes in uninsured rates (separated into commercial and other public program migration), high risk pool participation and characteristics, and others. Current commercial reinsurance levels for the individual market. The availability of rate filings to review current premium levels and rating parameters. Recommendation: We recommend that at least two simulations are run. The first will identify and allow for correction of data issues, while the second will be used to inform health plan pricing for To complete two rounds of simulations, work needs to begin very soon, no later than July The State should work to add further detail to the Urban Institute modeling and other analysis to help answer the key questions issuer actuaries will need to address for 2014 (specifically, further detail on Questions 2 and Question 3 in the prior list). 4.1 Different Approaches for Simulations Because simulations need to take place soon, and New York does not yet have a functioning APCD, there are two basic approaches to simulations: Approach 1: Issuers provide detailed data to state, state runs risk adjustment model on that data, and state (or their consultants) calculate risk adjustment results and distribute. Approach 2: Issuers run the risk adjustment model and provide summarized results to state and state (or their consultants) calculates risk adjustment results and distribute. 7 Page 17

20 The advantage to approach #2 is that it may be easier to get buy-in from issuers since they wouldn t have to submit detailed data. In addition, since the APCD will likely not be available for risk adjustment in 2014, a distributed approach for simulation would be consistent with the approach that will be used for implementation in However, approach #2 does not allow as much data validation because only summary results and some metrics would be reported. Recommendation: We recommend that the first risk adjustment simulation be a distributed approach. The second simulation should also be a distributed approach. However, as soon as the APCD is available, New York should run parallel testing against the distributed simulations to identify any limitations with the simulations, issues with APCD data or other related issues so that they can be corrected for Key Simulation Deliverables For risk adjustment, states should be able to create the following deliverables as a result of the simulation: Report to state entity governing risk adjustment illustrating the results of applying a risk assessment model on collected data from issuers. The results will include an average risk score for issuers which will allow the state entity to simulate risk adjustment payment calculations. Detailed results will include prevalence statistics and data diagnostics that will provide further insight into drivers of risk and data quality. Summarized relative risk scores will be shared with issuers providing critical input towards actuarial pricing of health insurance products in 2014 and Expected payable/receivable adjustment by carrier. This involves applying the risk score factor output mentioned above and normalizing for ratable factors (e.g., age). This also involves incorporation of baseline premium with possible adjustment for items such as geography and actuarial value of plan designs. Recommendation on whether or not to utilize a state model versus the federal model, and whether to administer the program locally. Explicit identification of expected deficit/surplus projected from risk adjustment, if expected, to be factored into the following year s contribution rate will need to be included in reports for late starting in late Publication of the risk adjustment model proposed to be used. The federal government proposes that states that plan to modify federal parameters (national in scope) issue their notice by early March in the calendar year before the effective date. Reinsurance output associated with the simulation would include the following: Page 18